Why are spinors not used more often in signal processing? To answer
that question, we have to find a way to explain what spinors are
really.

As far as i can see right now, the essence is about representing a
vector (a one--dimensional thingy) as a matrix. So, instead of looking
at the transform $Qx$, we look at $QXQ^{-1}$. This is a double cover
representation.

It brings us straight to Clifford algebras and higher order tensors.
For example, take a $16$--component vector. The spinor is the $Q$.  If
$X$ is a vector, $Q$ is a $128$--dimensional even grade element called
a spinor. One example of this is representing SO(3) as SU(2).

%% Now, can we do something with this? My immediate question would be,
%% what are the analogs of the matrix decompositions for this double cover
%% representation, and what can we say about displacement rank?
 
%% \section{Sat Oct 22 12:24:07 EDT 2005}

%% Been reading some subspace singal processing and identification stuff.
%% Something is should have done way earlier, really. Anyways. Books
%% arrived, so the tensor--sumsin estimation stuff will have to wait:
%% Papy and Delathauwer.